Does Right or Left Matter - Universidad Complutense de Madrid

V Complutense Internacional
Seminar On European Economy
Universidad Complutense de Madrid
Madrid, 10-11 May 2004
Does Right or Left Matter? Cabinets,
Credibility and Fiscal Adjustments
José TAVARES
Universidade Nova, Lisbon
Organised by:
European Economy Group
With the sponsorship of:
“Does Right or Left Matter?
Cabinets, Credibility and Fiscal Adjustments”*
José Tavares
Universidade Nova, Lisbon
JEL Classification Code: E62; H30; H62.
Keywords: Fiscal Adjustments; Cabinet Ideology; Credibility; Economic and Monetary Union;
Probit Analysis.
* José Tavares. Address: Faculdade de Economia, Universidade Nova de Lisboa, Campus de
Campolide,1099-032 Lisboa, Portugal. E-mail: jtavares@econ.ucla.edu
Abstract
This paper tests the widely-held assumption that left-wing cabinets favor higher public
spending, and examines whether cabinet ideology affects the persistence of major fiscal
adjustments. In a panel of large fiscal adjustments in OECD countries during the last 40 years, we
find evidence that left and right-wing cabinets are partisan: the left tends to reduce the deficit by
raising tax revenues while the right relies mostly on spending cuts. Our testable hypothesis is that
cabinets can thus signal commitment by undertaking fiscal adjustments in ways that are not favored
by their constituencies. In other words, the left gains credibility when it cuts spending while the
right becomes more credible when it raises tax revenues. Probit estimates of the determinants of
persistence in fiscal adjustments confirm that spending cuts by the left and tax increases by the
right are associated with persistent adjustments. The effect is significant for cuts in public spending,
public consumption (wage or non-wage), increases in total revenues, direct taxes on businesses and
other taxes. We test for the role of several other determinants of persistence, confirming that
coalition and majority cabinets are associated with less persistence while periods of high or rising
levels of indebtedness favor persistence. Estimates of the impact of ideology and composition of
the adjustment on the growth of GDP and its components shows that it is the size of the spending
cut rather than cabinet ideology that is most important.
2
1. Introduction
Economists have devoted a substantial effort to understanding the political economy of fiscal
adjustments, identifying several factors that favor the persistence of adjustments - including some
characteristics of the adjustment itself. Giavazzi and Pagano (1990) first uncovered cases of major fiscal
contractions associated with an expansion of output and consumption. Alesina and Perotti (1997) have
shown how adjustments that rely on spending cuts are more persistent and Perotti (1999) documented the
relationship between the economic context - periods of high deficits or high levels of public debt – and the
likelihood of success in a fiscal adjustment. When authors put forward an explanation for these empirical
regularities, the concept of “credibility” is almost always invoked.
This paper conducts an empirical search for credibility effects in fiscal adjustments. Specifically, we
test whether the match cabinet ideology – composition of the adjustment affects the probability of success.1
Our claim is that adjustments where left-wing cabinets cut spending or right wing cabinets increase taxes are
more likely to be successful.2 The reasoning is straightforward: since left and right-wing parties tend to
disagree over the ideal size of government3 and this disagreement is widely recognized by economic agents4
a cut in public spending by the left or an increase in taxes by the right signals commitment to the adjustment
and adds credibility to their actions. In sum, pursuing a policy that is painful to their own constituents,
cabinets signal the urgency of the fiscal adjustment. Our study is important since successful adjustments are
less contractionary than unsuccessful ones5 If we understand which factors increase the likelihood of
success, deficit cuts can be undertaken at a lower economic cost.
The paper is organized as follows. Section 2 briefly reviews the literature on fiscal adjustments and
presents our hypotheses. Section 3 presents the data set, documents partisanship during fiscal adjustments
and produces Probit estimates of the impact of different variables on the persistence of fiscal adjustments.
Section 4 concludes.
2. Fiscal Adjustments and Credibility
1
The terms success and persistence will be used interchangeably when applied to fiscal adjustments.
As suggested in the literature, center cabinets are seldom, if ever, successful when pursuing fiscal adjustments suggesting that
ideology may be an essential element in explaining success.
2
3
Another issue over which there may be partisan differences is the possible trade-off between unemployment and inflation, as in
Hibbs (1977) and Keech (1999). In this paper we do not address this issue, not least because the debate over whether there actually is
an unemployment-inflation tradeoff has been reopened.
4
As expressed by Robert Samuelson, when commenting on the 1996 Clinton-Dole Presidential campaign in the United States: ''(...)
the real issue concerns government direction. Will it grow bigger and more activist? Or will it be more restrained and disciplined?
And here genuine party differences exist.'' See Samuelson (1996).
3
In what circumstances can adjustments be expansionary? What is the role of credibility and of
ideological differences in particular? These are the two issues we will discuss in this section. In the standard
Keynesian framework, under the assumption of sticky wages in the short run, a cut in government spending
(or an increase in taxes) leads to decreases in output and consumption. The Keynesian framework is static
and only current policies matter. There is no role for expectations about the path of future policies and a
discussion of the issue of credibility does not make much sense. However, as ably pointed out in Bertola and
Drazen (1993) “a policy innovation that would be contractionary in a static model may be expansionary, if it
induces sufficiently strong expectations of future policy in the opposite direction''. A deficit cut can be
expansionary for several distinct reasons. As all public spending is ultimately financed by taxes (current or
future), a decrease in permanent spending implies a concomitant decrease in the present discounted value of
taxes. A credible decrease in permanent government spending lowers the net value of future tax receipts and
increases private wealth, leading to a boom.6 Increases in taxes can also lead to increases in output either
because it smoothes the path of taxes over time – when the distortionary costs of taxation depend nonlinearly on tax rates7 - or solve uncertainty over the future course of fiscal policy.8 Finally, cuts in the deficit
can raise household wealth - labor and non-labor - through a decrease in interest rates, as credible cuts in the
deficit can substantially lower interest rates, in tandem with expected inflation and default risk.
What has the literature uncovered as to the factors that make fiscal adjustments more credible? A
fiscal adjustment is defined as a change in the primary deficit above a given threshold,9 while success is
associated with the persistence of the deficit cut.10 The success of a fiscal adjustment depends crucially on its
credibility, i.e., how permanent the initial change in the deficit is believed to be. When a cut in the deficit is
perceived as permanent - i.e., unlikely to be reversed in the future – it may induce an expansionary effect
that contributes to its persistence through an increase in GDP that lowers the deficit as a share of GDP. The
literature on fiscal adjustments has put forward different factors that add credibility to the adjustment.
Giavazzi and Pagano (1990) suggested that the size of the deficit cut was key to the success of the Irish and
Danish adjustments of the mid-eighties.11 A second source of credibility is the composition of the deficit cut:
5
See Giavazzi and Pagano (1990), Alesina and Perotti (1997) and Alesina et al. (1998).
This is a corollary of the full discounting of future taxes behind the well-known Ricardian Equivalence result that purports that
changes in the deficit today do not lead to any change in consumption, interest rates or output if government spending is unchanged
and individuals discount future taxes appropriately. See Barro (1989).
7
This argument was originally made in Blanchard (1990).
8
For a given level of spending, tax increases have two effects on work effort: the wealth effect leads to increased work effort, higher
consumption and output; the substitution effect leads to reduced work effort, lower consumption and output. If the tax cut is
temporary, the substitution effect is likely to dominate, but if the tax increase is permanent, the wealth effect tends to dominate.
Pencavel (1986) shows that the substitution effect is likely to be small.
9
Usually defined as a change of 1.5 percent of GDP or more.
10
Defined as a total change over the two or three years after adjustment.
11
Giavazzi and Pagano (1996) found some evidence of an expansionary effect associated with larger adjustments. In contrast,
Alesina and Perotti (1997) find little association between size of the deficit cut and success.
6
4
Alesina and Perotti (1997) show that adjustments relying on spending cuts (rather than tax increases) tend to
be more successful. The public understands how these adjustments harm particular constituencies such as
trade union members and pensioners and thus assigns them a higher credibility.12 Finally, the circumstances
of the adjustment, in particular the level and increase of the public debt in the years before the adjustment
may further success. Perotti (1999) presents empirical evidence showing that, when public debt is high,
deficit cuts are more likely to be expansionary.13
An altogether different source of credibility is associated with the structure of the cabinet and the polity,
namely the number of “veto players” in the political system. Tsebelis (1995) defined veto players as
individuals or institutions (parties, the parliament, etc.) whose agreement is necessary to change the status
quo. The likelihood of policy change (e.g., fiscal adjustment) decreases as the number of veto players
increases.14 Coalition cabinets and minority cabinets increase the number of veto players in the system
relative to one-party majority cabinets, so that the credibility of a policy change such as a fiscal adjustment
can decrease.15 Empirically, Roubini and Sachs (1989) and Grilli et al. (1991) show that coalition cabinets
tend to be more fiscally irresponsible than other cabinets, increasing expenditures at the cost of larger
deficits. Volkerink and De Haan (2000) find that measures of political fragmentation, including the number
of parties in the cabinet, correlate positively with higher deficits.16
12
Alesina, Perotti and Tavares (1998) showed that the electoral success of a cabinet is not affected by the size of the cut. If anything,
larger deficit cuts and cuts in government wages increase the probability of survival of the cabinet. The reason why cabinets do not
adjust more often has to be that governments are sympathetic to or afraid of particular interest groups. See De Haan (1997) for a
reassessment of these findings.
13
Sutherland (1995) shows that, in a model where consumers have finite lives, a decrease in spending (or increase in taxes) has
contractionary effects at low debt levels but are expansionary at high debt levels. Bertola and Drazen (1993) also present a model
where cuts in spending have different effects at different levels of public spending but they do not derive clear empirical
implications.
14
Tsebelis (1995) highlights that less congruence – dissimilarity of positions between veto players – or less coherence – dissimilarity
of positions among elements of the same veto player institution – decrease the likelihood of policy change. Tsebelis (1999) shows
that political systems with more veto players and less congruence changed labor legislation less frequently while Tsebelis and Chang
(2001) find that an increase in the number of veto players in advanced industrialized countries leads to less frequent changes in the
composition of government budgets.
15
Alesina and Drazen (1991) and Spolaore (1993) have shown how coalition cabinets are less likely to reach agreement in such
matters as when and how to cut a government deficit. A coalition government has more difficulty than a single party majority
government to assign the costs of adjustment to specific groups since coalition partners are likely to veto some of the possible
options. One can extend the theoretical argument to argue that coalition cabinets are more likely to break an agreement over a fiscal
adjustment, decreasing its persistence.
16
In a reassessment of the results in Roubini and Sachs (1989), Edin and Ohlsson (1991) argue that it is minority cabinets that are
fiscally loose and not majority coalition cabinets, concluding that it is harder to negotiate with parliament than within the cabinet.
These authors find coalition cabinets to be as fiscally loose as single party majority cabinets, which suggests both the number of
parties in the cabinet and their support in parliament may be relevant. See also De Haan (1997) for a reassessment of the Roubini and
Sachs findings.
5
The success of fiscal adjustments might depend on the exchange rate regime in place.17 However,
Lambertini and Tavares (2000) examine in detail the role of exchange rates before, during and after fiscal
adjustments, and find a strong and robust association between a depreciation in the exchange rate before the
fiscal adjustment and the persistence of the ensuing adjustment, but the compositional effect and the other
determinants of persistence remain as important. Lane and Perotti (2001) also provide theory and evidence
on the role of exchange rate regimes but find their results are entirely consistent with the compositional
effects of fiscal policy.
Cabinets that undertake fiscal adjustments can reduce public spending, increase taxes or pursue a
mix of policies where at least one of the former actions is taken. One of the characteristics of cabinets that is
public is their ideological persuasion. Thus, if a left-wing cabinet adjusts by increasing taxes it may be
perceived as partisan, whereas if the same cabinet cuts spending it signals to economic agents the urgency
and benefits of the fiscal adjustment. This is one of the implications of the “why it takes a Nixon to go to
China” argument, formalized in Cukierman and Tommasi (1998): “when voters are not fully informed about
the way in which policies map into outcomes, policy proposals convey information.” In other words, policies
that benefit the majority of economic agents (such as deficit reduction) may be more easily implemented by
“unlikely” characters.18 While Drazen and Masson (1994) distinguish between the credibility of policies
versus that of policymakers, Cukierman and Tommasi (1998) argue, as we do in the current paper, that it is
the policymaker-policy pair that matters.
A key element is the motivation of politicians. In spite of the widely held prior belief that fiscal
adjustments are politically costly, Alesina at al. (1998) have found no evidence of that political cost in a
recent examination of poll data and cabinet turnover after fiscal adjustments. According to these authors,
cabinets do not bear an electoral cost after fiscal adjustments but avoid them due to risk aversion or to
protect special interests. Policy-makers that are simultaneously motivated by partisanship, the benefits of
office and aggregate welfare may, in circumstances such as a high public debt situation and in the absence of
electoral punishment, sacrifice partisan interests. Specifically, a left-wing cabinet may cut public spending –
if it buys them a higher likelihood of remaining in office.
In sum, our search for the effects of credibility on fiscal adjustments relies on four elements. First,
partisanship: policy-makers are ideological, i.e. the public recognizes the preferred course of action for each
17
More generally, the institutionalist view of Hallerberg and Von Hagen (1999) contends that it is the presence of budgetary
institutions, such as a strong finance minister and negotiated budget targets, that explains fiscal discipline.
6
cabinet. Second, asymmetric information: the economic benefits of reducing the deficit are uncertain and the
cabinet has better information than the public. Third, signaling: by pursuing policies that are costly and
against the immediate interests of their constituents, policy-makers provide information on their economic
benefits. Fourth, feedback: signaling policy commitment improves the chances of the fiscal adjustment being
successful.
3. Empirical Analysis
3.1. Data and Summary Statistics
Our data set covers nineteen countries in the OECD between the years 1960 and 199519 and relies on
two main sources. Macroeconomic indicators - including the fiscal policy variables - are from the OECD
Economic Outlook (OECD (1997)) while cabinet characteristics are taken from Budge et al. (1993), updated
to 1995 by Woldendorp et al. (1998). The Political Handbook of the World (1996) was used by the author to
compute political data for Greece, Portugal and Spain, as well as cabinet changes for 1995, using the criteria
in Budge et al. (1993).20 The ideological index in Budge et al. (1993) locates all cabinets on a left-right scale
according to the share of seats in government (or of supporting parties in parliament when the latter
explicitly support the cabinet).21 This index takes values between 1 and 5, taking the value 1 if the share of
right-wing seats is equal to or greater than 2/3, 2 if it is between 1/3 and 2/3, 3 if the center parties account
for more than 50 percent of seats (or left and right-wing parties balance each other out in the cabinet), with
the values 4 and 5 for left-wing dominated cabinets likewise defined. We also use three alternative indices of
cabinet ideology, as defined in Appendix I and in the text below. Other cabinet characteristics are the
number of parties and whether the cabinet holds a majority. If more than one party holds posts in the cabinet,
the cabinet is classified as a coalition; if the parties supporting the cabinet hold a majority of seats in the
lower house of Parliament, it is a majority cabinet. In sum, cabinets are classified in a right-left scale (taking
values between 1 and 5, where 5 is most leftist) and according to whether they are majority or coalition
cabinets (with each characteristic noted by a dummy variable).22
18
Rodrik (1993) pointed out that dramatic fiscal, trade or institutional reforms are carried out by populist governments. Cukierman
and Tommasi (1998) carry the argument further and suggest that those parties may actually have an advantage over other parties
when carrying out those reforms.
19
The countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the
Netherlands, Norway, Portugal, Spain, Sweden, the United Kingdom and the United States.
20
Greece, Portugal and Spain are included only from appropriate dates in the mid-1970s, when these countries became democratic.
21
The classification of a party´s ideological position used by Budge et al. (1993) is taken from Castles and Mair (1984), a crossnational study that surveys experts, mostly political scientists, on party ideology. Castles and Mair (1984) place each party on a
segment representing left-right political orientation with the aim of arriving at a political scale that is comparable across countries.
22
The cabinet considered is that which is in power for most of the calendar year according to Budge et al. (1993).
7
In Table 1 below we present summary statistics for the main variables during fiscal adjustments. We
define a fiscal adjustment as a year for which the change in the primary deficit is –1.5 or less. An adjustment
is successful if the total change in the primary deficit in the three years after the tight period is –1 percent (or
less) of GDP or, three years after the initial adjustment year, the debt to GDP ratio is 5 percent below its
level before the tight period.23 An analysis of the distribution of fiscal adjustments by country and year
shows that they are well distributed in both dimensions. Table 1 shows that the size of the deficit cut is not
radically different for successful and unsuccessful adjustments, confirming previous results in the literature.
However, there is a strong composition effect with successful adjustments reducing spending more than
increasing taxes and unsuccessful adjustments doing the opposite. Coalition and majority cabinets are both
less likely to be successful in fiscal adjustments and the level and the change in public debt are higher just
before successful adjustments. There are no major differences as to the stage of the cycle at which both types
of adjustments are undertaken, even if successful adjustments tend to occur when unemployment is
decreasing rather than increasing.
[Table 1 about here]
3.2. Ideology and Fiscal Policy
To motivate the argument that left and right-wing cabinets might differ in fiscal adjustments we
present three pieces of evidence: opinion surveys, the empirical literature on ideology and fiscal policy and
statistics computed in our sample. For our purposes, it suffices to determine whether the preferred size of
government differs for left and right and that this difference has fiscal consequences.24 Several surveys
document the association between ideology and fiscal preferences. Kay et al. (1995) reports results from a
broad sample of interviews in 1992 and 1995. Individuals were asked to locate themselves in both the
Republican-Independent-Democrat and the Conservative-Moderate-Liberal spectra and then asked the
question: “In order to reduce the deficit, what is your desired percentage decrease in spending for each
point increase in taxes?”25 The survey results are clear: Republicans favor larger spending cuts than
Democrats for each point of increase in taxes, 1.98 instead of 1.43 and 2.95 instead of 1.75, respectively in
1992 and 1995. This difference of 0.55 and 1.10 percent of GDP, respectively for 1992 and 1995, is not
23
This is virtually the same definition as used in Alesina et al. (1998). These and other authors have experimented with other
definitions of fiscal adjustment and find that the empirical results change little.
24
We will not discuss the origin of the different character of left and right - discussed, for instance, in Lewis-Beck (1988) - nor
where the diverging preferences for government size come from. The obvious candidate is that parties cater to different
constituencies and parties on the left cater to a lower income constituency who benefit more from public expenditure programs
and/or faces lower tax rates. Piketty (1995) quotes survey evidence associating income and voter preferences that supports the
contention that lower income individuals tend to vote with the left.
25
The question mentions explicitly tax increases and spending cuts and thus captures precisely what we are interested in: the
individual trade-off between the benefits of public spending and the costs of taxation. Hansen (1998) states that “[c]itizen’s
8
negligible. The difference between Conservative and Liberals is similar, at 0.87 and 0.99 for the same two
years.26 Other surveys have found similar results. Gallup (1990) gauged individual support for senators that
had favored tax increases. Individuals who classified themselves as Democrats (or Liberals)27 were more
likely to grant their support than those defining themselves as Republicans (or Conservative) and the
difference was larger than for other individual characteristics - such as gender, age, race and education.
Hansen (1999) finds that individuals identified with the left favor higher taxes and higher spending.
Interestingly, no similar partisan differences can be uncovered with respect to the deficit. Hansen (1998)
documents how most of the voters prefer the status quo when choosing between greater domestic spending,
lower deficits and lower taxes. Consistent with the finding in Blinder and Holtz-Eakin (1984) that individual
ideology has no direct effect on support for a balanced-budget amendment.
The literature that tests the link between cabinet ideology and fiscal actions reveals evidence in favor
of the hypothesis that leftist cabinets increase spending more than rightist cabinets. Such is the case in
Golden and Poterba (1980) – for the US between 1952 and 1978 – and Roubini and Sachs (1989) – for the
OECD.28 In contrast, and despite the popular assumption that left-wing cabinets are more prone to deficit
spending, empirical studies such as Hahm, Kamle and Mowery (1995) have been unable to uncover a clear
correlation between partisanship and the size of the deficit
We now turn to our data set and classify cabinets into left and right according to four different
indices, as mentioned in Appendix I. Our Ideology index, which will be used throughout the empirical work,
is complemented by three additional indices, made available by George Tsebelis from the original Castles
and Mair (1984), Laver and Hunt (1992) and Warwick (1994) articles. Warwick (1994) generates its
ideological index from forty different measures presented from experts, party manifestos and survey sources.
Castles and Mair (1984) generate their ideological scores from a questionnaire survey of more than 115
political scientists in Western Europe and the United States. Each expert was asked to place the parties
holding seats in the national legislature on a left-right political spectrum ranging from 0 (extreme left) to 10
(extreme right), with 2.5,5 and 7.5 representing the moderate left, the center and the moderate right,
preferences over the public budget are remarkably well structured. Most people have no difficulty in determining their views. Most
have no difficulty recognizing tradeoffs”. (Italics added)
26
It is remarkable that middle of the road voters (Independents/Moderates) display middle of the road fiscal preferences, i.e., they
favor spending cuts that are larger than those favored by the left, but smaller than those favored by the right. These results suggest
that preferences over the level of public spending are systematically associated with preferences on a broader set of political issues
that determine individual political alignment.
27
The use of the two dimensions attempts to correct for prejudices associated with specific labels that could bias results. However,
here as in other studies, the results do not depend in any way on the labels used.
28
The study of the components of public spending does not uncover dramatic differences either. Blais et al. (1993), for instance,
conclude that “parties do make a difference, but a small one”.
9
respectively.29 Finally, Laver and Hunt (1992) asked experts to locate the policy positions of party leaders
and party voters according to whether they favored relatively more an increase in public services or a cut in
taxes.30 The four alternative indices have varying sample sizes with our Ideology index covering the most
countries and cabinets. In addition, our Ideology index displays a simple correlation of 0.69, 0.56 and 0.75
with each of the other three indices, respectively Ideology LH, Ideology CM and Ideology W.
The first evidence of cabinet partisanship is the fact that their behavior is different in adjustment
years than in other years. In addition, while changes in spending and changes in revenue display a positive
correlation of 34 percent in the sample as a whole, this correlation more than doubles to 73 percent during
adjustments. In other words, during fiscal adjustments, cabinets tend to increase or decrease spending and
revenues simultaneously. If these are adjustment years, this may be evidence that they tend to be partisan,
i.e. decrease the deficit by changing government size towards their preferred ideal level.31 Table 2 presents
strong evidence of partisanship during adjustment years. We report the change in the deficit, public spending
and revenues for the sub-samples of left and right cabinets, during years of fiscal adjustments. We find that,
regardless of the Ideology index used, the average change in the deficit is remarkably similar between left,
right and center cabinets, while left and right differ substantially as to the average composition of the deficit
cut. If on the left, cabinets tend to increase taxes more than cut expenditures while, if on the right, the
opposite holds. More than half of the cut in the deficit by the left relies on tax increases, while the opposite
holds for right-wing cabinets.32 This, we believe, is in accordance with the prior belief held by most
economists and policy-makers. Partisanship during fiscal adjustments is also suggested by a comparison of
the simple correlation between ideology and fiscal changes. While all indices of cabinet ideology have a
correlation coefficient very close to 0 in the whole sample, during years of fiscal adjustments, all ideological
indices with the exception of Ideology W display coefficients of correlation of 17 percent or more.33 This
evidence suggests that cabinets tend to be partisan during adjustments, implying that they can gain
credibility when they are not partisan and adjust in ways contrary to the interests of their constituents.
[Table 2 about here]
3.3. Specification and Estimates
29
The authors present results only for countries where at least three experts provided answers and the ideological score was the
average of all available responses.
30
Respondents assigned each party a score between one (“favors raising taxes to increase public services”) and 20 (“favors cutting
public services to cut taxes”).
31
So that, for instance, left-wing cabinets might decrease the deficit by increasing public spending and increasing taxes at the same
time.
32
Most of this difference between left and right is accounted for by government wages and non-wage consumption and transfers.
33
The positive sign of the correlation suggests that left-wing cabinets (assigned a higher ideology index) tend to promote positive
changes in spending and taxes, especially during adjustments, suggesting a partisan choice.
10
Previous studies and Table 1 above show that adjustments relying on spending cuts are the most
successful – the compositional effect. Table 2 shows that left and right tend to be partisan during fiscal
adjustments, that is, the left is more likely to increase taxes and the right cut spending. These two facts lead
us to expect the right to be much more successful than the left in cutting the deficit. That is not the case. As
Table 1 documents, confirming previous results in the literature, left-wing cabinets constitute virtually the
same share of adjustments, whether successful or unsuccessful. The only way to reconcile the three facts –
successful adjustments rely on spending cuts, the left cuts spending less than the right but is not less
successful for that - is for the success rate of left and right-wing to differ when pursuing adjustments in
different ways. Specifically, it must be the case that, though the left is less likely to cut spending, when it
does it is more successful than the right.34 This suggests the hypotheses to be tested empirically: Hypothesis
1 - An adjustment pursued by cutting public expenditure is more likely to be successful if pursued by the left
than by the right; Hypothesis 2 - An adjustment pursued by increasing the level of taxes is more likely to be
successful if pursued by the right than if it is pursued by the left.
Our basic specification estimates the effect of the different sources of credibility on the probability
of a fiscal adjustment being successful. We compute Probit estimates of the impact of the independent
variables on the probability of success for the sample of adjustment years. The dependent variable takes the
value 1 if the adjustment is successful and 0 otherwise.35 Our main variable of interest is the interaction of
ideology and the change in the fiscal variable. In addition to their interaction, we add the ideology indicator
and the change in fiscal variable as independent controls.36 If the coefficient associated with the ideologyfiscal change interaction is significant, we interpret it as evidence of credibility effects associated with
cabinet ideology. An analysis of the results for different components of spending and revenues will allow us
to determine which sub-categories of spending and revenues are the important sources of credibility. We
include other controls in the basic specification, including the type of cabinet initiating the adjustment
(whether it is a coalition cabinet and it holds a majority), the stage of the business cycle (proxied by the rate
of growth of real GDP and of the unemployment rate) and the fiscal position at the time of the adjustment
(level of public debt and change in public debt before the adjustment year).
It is important to mention the controls we do not include. The size of the adjustment is omitted since
there is a high correlation between changes in the deficit and changes in spending and its components, but
34
Conversely, the right is more successful in the few instances when it increases taxes.
As mentioned above, a fiscal adjustment is defined as a yearly change in the deficit of at least -1.5 percent of GDP. It is considered
successful if the decrease in the average deficit is at least 1 percent of GDP three years after the adjustment year or the level of
public debt decreases by 5 percent after those three years. The dummy for success takes the value 0 when neither of the criteria are
met.
35
11
our results on the ideology indicators are robust to the addition of the size of the deficit cut. More
importantly, we do not add indicators of the election cycle. These could be important in light of partisan
political business cycle theory, which proposes a relationship between economic policy and the electoral
cycle, as in Hibbs (1977) and surveyed in Alesina et al. (1997), which extend the discussion to rational
partisan cycles. Two reasons support our exclusion of electoral dating variables: first, empirical studies have
found little evidence of significant effects of election dates on fiscal policy,37 and we have confirmed that our
results are robust to inclusion of these variables; second, in our sample we find no evidence that major fiscal
adjustments tend to occur at a particular point in the electoral cycle or be more successful when they do.38
One way to deal with reverse causation is the fact that the dependent variable depends on
information available three years into the future – the success indicator - whereas the independent variables
are measured at the time or before the adjustment. In Appendix II we present sample summary statistics that
corroborate the lack of reverse causation. The first table presents the share of observations with cabinets that
have been in power for one year and up to four years. We examine the whole sample, the years of fiscal
adjustment and the years of fiscal adjustments classified as success, and find that the distribution of cabinets
across tenure years is virtually unchanged across the samples. The typical adjustment - successful or not occurs in the second year of tenure. When we examine the number of changes in the cabinet ideology (due to
an election or parliamentary reshuffles), again we do not discern any pattern of differences across samples.
There is no substantially higher likelihood of a leftward or rightward change before adjustments, whether
successful or not. Thus, the timing of adjustments and the probability of an ideology change just before the
adjustment suggest that fiscal adjustments are not undertaken directly in response to the electorate. We can
interpret the results below as documenting a relationship running from fiscal changes to success and not the
reverse.
In Tables 3 and 4, we present specifications for different independent variables, each corresponding
to the interaction of the ideology index with the change in a specific fiscal variable, shown at the top of each
column. A negative sign in the coefficient associated with this interaction indicates that a cut in the spending
category by cabinets further to the left is associated with a higher likelihood of success. When we analyze
the results for the different components of public spending in Table 3, we verify that total public spending
and public consumption – both its wage and non-wage components – are all significant at the 10 percent
36
Thus, we are able to control for the role of ideology in success, independently of the fiscal action undertaken. And likewise for the
effect of composition of fiscal change, independently of the cabinet ideology.
37
See Alesina et al. (1997).
38
See Appendix II, which presents summary statistics on the timing of fiscal adjustments, the timing of changes in the ideology of
the cabinet and success of the adjustment. There is no clear relationship between the time at which there are cabinet changes, the
ideological changes involved and the pursuit or success rate of adjustments.
12
level or above. The estimated size of the coefficients is higher for public consumption and its two
components, suggesting a dramatic increase in the credibility for an adjustment by a left-wing cabinet that
cuts public spending. The estimate of -7.92 for the coefficient on the “cabinet ideology – change in public
spending” pair suggests that a 1 percent cut in spending by the leftmost cabinet has a 32 percent higher
likelihood of success than the exact same policy undertaken by the rightmost cabinet. This is an important
difference. Notice that the Ideology indicator suggests that left-wing cabinets are less likely to be successful
but the size of the coefficient is very small, so that when pursuing the credible policy, left-wing cabinets
easily attain more success. As to the variable capturing the change in fiscal policy in isolation, it is never
even close to significance, except in the case of public investment when the coefficient has the sign opposite
to what is expected. As to the control variables, both the characteristics of the cabinet and the fiscal status
have strong effects on the likelihood of success, so that increases in the level of public debt, and minority
single party cabinets tend to be more successful. The type of government has a noticeable effect on the
likelihood of a persistent adjustment: other things being equal, as an adjustment pursued by a coalition or
minority cabinet is around 30 percent less likely to be persistent.39
[Table 3 about Here]
Table 4 presents results for changes in total tax revenues and its components. The results are similar
to those for spending in that the cabinet type and the recent change in fiscal position matter for success. As to
our main variable of interest, Ideology*Change in Revenues, it tends to be associated with a negative
coefficient,40 suggesting that increases in tax revenues by the left are less likely to lead to persistent
adjustments. The coefficient is significantly different from 0 in the case of Other Taxes and Direct Taxes on
Business, which are significant at the 1 percent level. These conditional coefficients suggests that a 0.25
percent increase in other tax revenues – the average for the sample of adjustments - leads to about 34 percent
higher rate of success for the most left-wing cabinet as compared to the most right-wing. Note that, while the
coefficient for the change in revenues in isolation is significant in both the Other Taxes and the Direct Taxes
on Business, this sign is opposite to what would be expected if only the change mattered for credibility.
[Table 4 about Here]
39
Why minority single party cabinets are more successful than others can be due to a series of factors. First, using the terminology in
Tsebelis and Chang (2001), a single party cabinet has more coherence. The fact that the cabinet does not hold a majority may
decrease congruence, i.e. similarity of positions, between veto players. However, three other forces may lead to more congruence:
first, before fiscal adjustments the fiscal situation tends to deteriorating so rapidly that agreement becomes easier to attain; secondly,
fiscal changes that are credible represent a compromise among veto players – left-wing cabinets decrease spending or right-wing
cabinets increase taxes – so that agreement is also made easier; third, the impact of the adjustment on the economy, which is likely to
be positive in successful adjustments, also adds support for the adjustment, independently of how it is to be achieved.
40
The exception is indirect taxes.
13
Table 5 tests whether our results are robust to different ideology coefficients, showing that
the interaction of the Ideology indices and changes in the fiscal variable all come out negative, suggesting a
credibility effect. While all but one of the interacted variables related to changes in spending are strongly
significant, the interaction of ideology and changes in revenues displays lower significance. We ran the same
specifications for the different components of spending and revenues - as in Tables 3 and 4 – for the three
additional ideology indices and found that the variables that were significant in those tables remained
significant, namely Public Consumption and its components (Wage and Non-wage ), Other Taxes and Direct
Taxes on Business.
[Table 5 about Here]
An issue requiring our attention is the extent to which the above results would change if cycle
corrected fiscal variables were used instead. We have used cyclically adjusted definitions of the deficit to
define the years of fiscal adjustments and our results remain unchanged. Moreover, the fact that business
cycle indicators are never significant determinants of the timing or success of fiscal adjustments in Table 4
suggests that business cycle fluctuations do not explain our results.41
An important question is whether credibility originates different reactions to fiscal policy changes as
far as GDP and its components are concerned. We start to address this question by considering that, if an
adjustment is credible, variables such as output, investment and consumption may respond positively to the
deficit cut. In Table 6 we test for the effect of the ideology-fiscal policy interaction for the case of public
spending and its most important component, public consumption. For reasons of parsimony we report only
the coefficients on the ideology-fiscal policy interaction and the two variables in isolation, but we have used
all controls as in Tables 3 through 5. We find that it is the change in fiscal policy, not the interaction with
ideology, that matters for output, investment and consumption growth.42 In fact, the negative sign on the
coefficient of the fiscal policy change is statistically significant and suggests the larger the decrease in public
spending or public consumption, the larger the increase in output and its components. This increase is
substantial, with the share of business investment in GDP increasing between 7 and 14 percent in the two
years after the adjustment. This is the first time the composition effect is tested for in terms of impact on
output.
41
Other papers in the literature, such as Alesina et al. (1998), have found no change in results when fiscal variables corrected for the
cycle were used instead of simple shares of GDP. Hallerberg and Strauch (2001) have found extremely small cyclical variation in
spending and tax receipts as shares of GDP for the European countries in our study.
42
Here, as in Tables 3 through 5, the correlation between the interaction variable and the change in fiscal policy variable is high.
14
[Table 6 about Here]
4. Conclusion
This paper draws on the assumption that left-wing cabinets favor a larger size of government than
right-wing cabinets to assess the role of credibility in the persistence of fiscal adjustments. The literature on
fiscal adjustments has documented two important facts on the determinants of success:43 First, fiscal
adjustments that rely on spending cuts are more persistent than those relying on tax increases; second, fiscal
adjustments initiated by left-wing cabinets are as persistent as those initiated by right-wing cabinets. In this
paper we document a third fact: when a left-wing cabinet cuts the deficit it tends to rely on tax increases,
whereas the right tends to rely on expenditure cuts. For all three facts to be internally consistent, it must be
that left and right-wing cabinets have different success rates when they cut the deficits in different ways, that
is, despite the fact that the left tends not to cut spending during adjusting, when it does, it gains in credibility.
Using data for a large sample of OECD countries, we show that it is during fiscal adjustments that
the difference between the fiscal actions of left and right-wing cabinets is clearer, with the left relying mostly
on tax increases and the right on spending cuts. We estimate how the interaction of cabinet ideology and
fiscal change affect the likelihood that the adjustment is persistent. We find evidence in support of the
hypothesis that the left is more credible than the right when cutting expenditures – total expenditures, public
consumption and its components – and the right more credible when it increases taxes – namely other taxes
and direct taxes on businesses. We then investigate whether fiscal policy actions by left and right-wing
cabinets have differential impacts on output, private investment and consumption immediately after the
adjustment. Our evidence shows that it is the fiscal policy change in itself that matters for the postadjustment boom in output and its components. These results are consistent with the literature on
composition of adjustment but we go beyond the literature by explicitly testing for credibility effects on the
persistence of adjustment and the determinants of the post-adjustment boom. Our paper provides strong
evidence that, as far as the persistence of fiscal adjustments is concerned, what is important is not only what
you do, it is how you do it and who you are.
43
As documented in Alesina and Perotti (1997) and Alesina, Perotti and Tavares (1998).
15
APPENDIX I – THE DATA
CABINET DATA
Ideology - The cabinet ideological index developed by Budge et al. (1993) and updated by Woldendorp et
al. (1998) locates the cabinet on a left-right scale, taking the values 1 to 5. It has been extended to include
Greece, Portugal and Spain (from the respective democratic transitions) by the author using the Political
Handbook of the World (1996). The paper uses three alternative indices of cabinet ideology in addition to
the previous one, based respectively on Castles and Mair (1984), Laver and Hunt’s (1992) and Warwick
(1994) and named Ideology CM, Ideology LH and Ideology W. These last three indices were normalized so
that they varied from a minimum value of 1 and a maximum of 5, with the latter denoting most left-wing.
Coalition – A dummy variable taking the value 1 if a coalition cabinet - which includes ministers from two
or more parties- is in power.
Majority – A dummy variable that takes the value 1 if the cabinet has majority support in parliament.
MACROECONOMIC DATA
Change in Deficit – The yearly change in the primary deficit as a share of Gross Domestic Product. All
other fiscal variables are defined in a similar way: Change in Spending and Change in Revenues are,
respectively, the changes in primary public expenditures and in public revenues as a share of GDP. We also
use the disaggregated fiscal policy variables Change in Public Consumption, Change in Public Investment,
Change in Transfers and Change in Subsidies, while Change in Public Wage Expenditure and Change in
Non-wage Public Expenditure sum up to Change in Public Consumption. Change in Direct Taxes, Change
in Indirect Taxes, Change in Other Taxes are used, in addition to Change indirect Taxes On Households and
Change in Direct Taxes on Businesses sum up to the total change in direct taxes. We also use the Change in
Social Security Payments and Change in Social Security Revenues as dependent variables.
Lagged Debt – Level of public debt as a share of Gross Domestic Product at the beginning of the fiscal
year, in percent. The Lagged Change in Debt is the change in the level of public debt as a share of Gross
Domestic Product in the last three years before the first adjustment year.
Growth GDP – Rate of growth in real GDP between the current year and the year before, in percent.
Growth UNR – Difference between the unemployment rate in the current year and the year before, divided
by the unemployment rate the year before, in percent.
Growth PI - The growth rate of private investment as a share of GDP, in percent. Growth HI is similarly
defined for investment in housing and Growth BI for business investment, their sum adding up to Growth
PI. Growth PC is similarly defined for private consumption.
16
Appendix II – Ideology, the Timing and Success of Fiscal Adjustments
The Timing of Fiscal Adjustments and Success
Year 1
All Observations
Fiscal Adjustments
Fiscal Adjustments and Success
Year of Tenure for Cabinet
Year 2
Year 3
Year 4
0.22
0.19
0.36
0.31
0.21
0.22
0.31
0.20
0.20
0.13
0.11
0.07
Note: A fiscal adjustment is a change in the primary deficit by at least -1.5 percent of GDP in a single year. Success is defined as a decrease
in the fiscal deficit in the three years following the adjustment.
Changes in Ideology, Fiscal Adjustments and Success
All Observations
Fiscal Adjustments
Fiscal Adjustments and Success
Change in Ideology Year Before
No Change
To the Left
To the Right
731
65
50
82
7
8
41
5
6
Note: A fiscal adjustment is a change in the primary deficit by at least -1.5 percent of GDP in a single year. Success is defined as a decrease
in the fiscal deficit in the three years following the adjustment.
17
Acknowledgements: The author has benefited from comments by Alberto Alesina, David Baron, Mário
Centeno, David Cutler, Jeffry Frieden, Tim Groseclose, Fernanda Llussá, Francisco Rodriguez, Dani
Rodrik, George Tsebelis, Romain Wacziarg, two anonymous referees and the editor, as well as participants
at seminars at Stanford University, University of California at Los Angeles, Universidade Nova and the VIII
World Conference of the Econometrics Society, Seattle 2000. Roberto Perotti and George Tsebelis have
kindly provided part of the data used in the paper. Financial support from the Banco de Portugal, the
Fundação para Ciência e Tecnologia and INOVA is gratefully acknowledged. Any remaining error is my
own.
18
References
Alesina, A., Drazen, A., 1991. Why are Stabilizations Delayed? American Economic Review 81(5), 11701188.
Alesina, A., Perotti, R., 1997. Fiscal Adjustments in OECD Countries: Composition and Macroeconomic
Effects. International Monetary Fund Staff Papers 44(2), 210-248.
Alesina A., Roubini, N., Cohen, G., Political Cycles and the Macroeconomy. MIT Press, Cambridge,
Massachusetts and London.
Alesina A., Perotti, R., Tavares, J., 1998. The Political Economy of Fiscal Adjustments. Brookings Papers
on Economic Activity 0(1), 197-248.
Barro, R., 1989. The Neoclassical Approach to Fiscal Policy. In: Barro, R., (ed.), Modern Business Cycle
Theory, Harvard University Press, Cambridge, Massachusetts, pp. 178-235.
Bertola, G., Drazen, A., 1993. Trigger Points and Budget Cuts: Explaining the Effects of Fiscal Austerity.
American Economic Review 83(5), 1184-1198.
Blais, A., Blake, D., Dion, S., 1993. Do Parties Make a Difference? Parties and the Size of Government in
Liberal Democracies. American Journal of Political Science 37, 40-62.
Blanchard, O., 1990. Comments on Giavazzi and Pagano. In: Blanchard, O., Fischer, S., (eds.), NBER
Macroeconomics Annual 1990, Cambridge, MA, MIT Press, pp. 112-117.
Blinder, A., Holtz-Eakin, D., 1984. Public Opinion and the Balanced Budget. American Economic Review
74(2), 144-149.
Budge, I., Kemam, H., Woldendrop, J., 1993. Political Data 1945-90. European Journal of Political
Research, Special Issue 24, 1-120.
Castles, F., Mair, P., 1984. Left-Right Political Scales: Some “Expert Judgements. European Journal of
Political Research 12, 73-88.
Cukierman, A., Tommasi, M., 1998. When Does It Take a Nixon To Go To China. American Economic
Review 88(1), 180-197.
De Haan, J., 1997. Political and Economic Determinants of Budget Deficits and of Government Spending: A
Reinvestigation. European Journal of Political Economy, 13(4), 739-750.
Drazen, A., Masson, P., 1994. Credibility of Policy versus Credibility of Policymakers. Quarterly Journal of
Economics, August, 735-754.
Fuller, D., Alston, R., Vaughan, M., 1995. The Split Between Political Parties on Economic Issues: A
Survey of Republicans, Democrats, and Economists. Eastern Economic Journal 21, 227-238.
Gallup Organization, 1990. The Gallup Poll Monthly. May.
19
Giavazzi, F., Pagano, M., 1990. Can Severe Fiscal Adjustment be Expansionary? Tales of Two Small
European Countries. In: Blanchard, O., Fischer, S., (eds.), NBER Macroeconomics Annual 1990, pp. 75111, MIT Press, Cambridge, Massachusetts.
Giavazzi, F., Pagano, M., 1996. Non-Keynesian Effects of Fiscal Policy Changes: International Evidence
and the Swedish Experience. Swedish Economic Policy Review 3(1), 67-103.
Golden, D., Poterba, J., 1980. The Price of Popularity: The Political Business Cycle Reexamined. American
Journal of Political Science 24, 696-714.
Grilli, V., Masciandaro, D., Tabellini, G., 1991. Political and Monetary Institutions and Public Financial
Policies in the Industrial Countries. Economic Policy: A European Forum 6 (2), 341-392.
Hahm, S., Kamlet, M., Mowery, D., 1995. Influences on Deficit Spending in Industrialized Countries.
Journal of Public Policy 15, 183-197.
Hallerberg, M., and Strauch, R., 2001. On the Cyclicality of Fiscal Policy in Europe. Mimeo, European
Central Bank, Frankfurt.
Hallerberg, M., and Von Hagen, R., 1999. Electoral Institutions, Cabinet Negotiations, and Budget Deficits
in the European Union. In Poterba, J., von Hagen, J., (eds.) Fiscal institutions and fiscal performance. NBER
Conference Report series. Chicago and London: University of Chicago Press, 1999, pages 209-32.
Hansen, J. M., 1998, Individuals, Institutions, and Public Preferences over Public Finance. American
Political Science Review 92 (3), 513-531.
Hansen, J. M., 1999. Thinking the Budget: Public Opinion and Tradeoffs Over Spending, Taxes and
Deficits. Mimeo. Department of Political Science, University of Chicago.
Hibbs, D., (1977). Political Parties and Macroeconomic Policy. American Political Science Review, 23,
1467-1488.
Kay, A., Henderson, H., Steeper, F., Greenberg, S., Blunt, C., 1995. Who Will Reconnect With the People:
Republicans, Democrats, or ... None of the Above? Americans Talk Issues, August, Survey #28.
Keech, W., 1995. Economic Politics: The Costs of Democracy. New York: Cambridge University Press.
Lambertini, L., Tavares, J., 2000. Exchange Rates and Fiscal Adjustments: Evidence From the OECD and
Implications for the EMU. Mimeo, University of California Los Angeles.
Lane, P., Perotti, R., 1998. The Trade Balance and Fiscal Policy in the OECD. European Economic Review,
42, 887-895.
Lane, P., Perotti, R., 2001. The Importance of Composition of Fiscal Policy: Evidence from Different
Exchange Rate Regimes. Mimeo, European University Institute.
Laver, M., Hunt, W. B., 1992. Policy and Party Competition. New York: Routledge, Chapman and Hall.
Organization of Economic Cooperation and Development, 1997. Economic Outlook. OECD, Paris.
Perotti, R., 1999. Fiscal Policy in Good Times and Bad. Quarterly Journal of Economics 114(4), 1399-1436.
20
Political Handbook of the World, 1996. McGraw-Hill, New York.
Rodrik, D., 1993. The Positive Economics of Policy Reform, American Economic Review 83, 356-361.
Roubini, N., Sachs, J., 1989. Political and Economic Determinants of Budget Deficits in the Industrial
Democracies. European Economic Review 33(5), 903-933.
Samuelson, R., 1996. The Debate We Need .… Newsweek, 7 October 1996.
Spolaore, E., 1993. Policy Making Systems and Economic Efficiency, Coalition Governments vs. Majority
Government. Unpublished paper, Harvard University.
Sutherland, A., 1997. Fiscal Crisis and Aggregate Demand: Can High Public Debt Reverse the Effects of
Fiscal Policy? Journal of Public Economics 65(2), 147-162.
Tsebelis, G., 1995. Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism,
Multicameralism and Multipartyism. British Journal of Political Science, 25, 289-325.
Tsebelis, G., 1999. Veto Players and Law Production in Parliamentary Democracies: An Empirical Analysis.
American Political Science Review, 93(3), 591-608.
Tsebelis, G., Chang, E., 2001. Veto Players and the Structure of Budgets in Advanced Industrialized
Countries. Mimeo, University of California Los Angeles.
Volkerink, B., De Haan, J., 2000. Fragmented Government Effects on Fiscal Policy: New Evidence. Mimeo.
Department of Economics, University of Groningen.
Warwick, P., 1994. Government Survival in Western European Parliamentary Democracies. New York:
Cambridge University Press.
Woldendorp, J., Keman, H. and Budge, I., 1998, “Party Government in 20 Democracies: an Update (1990–
1995)”, European Journal of Political Research 33: 125–164, 1998. 125.
21
Table 1
Characteristics of Adjustments, Successful and Unsuccessful
All Sample
Change in Deficit
Change in Spending
Change in Revenues
Lag Debt
Lag Change Debt
Growth UNR
Growth GDP
Majority
Coalition
Left
Right
0.12
0.57
0.46
44.53
2.93
1.29
3.46
73.44
50.42
32.50
49.17
Fiscal Adjustments Successful Unsuccessful
Adjustments Adjustments
-2.33
-2.56
-2.17
-1.05
-1.71
-0.58
1.29
0.87
1.59
54.39
68.63
44.00
7.57
11.18
4.92
-0.11
-3.44
1.70
3.42
3.42
3.41
71.72
57.14
79.69
49.49
34.29
57.81
38.38
37.14
39.06
42.42
60.00
32.81
Note: Change in Deficit, Change in Spending and Change in Revenues correspond, respectively, to the change in the primary public deficit, public
expenditures and government revenues as a share of Gross Domestic Product (GDP), in percentage points. Lag Debt and Lag Change Debt are the
level of public debt at the time of adjustment and the change in its level in the 3 years before the adjustment, as shares of GDP, in percentage points.
The variables Growth Unemployment and Growth GDP are the yearly change in the unemployment rate and in real GDP, in percentage points.
Majority, Coalition, Left and Right give the frequency of cabinets with each characteristic in the corresponding sample. See Appendix I for the
description of the variables.
22
Table 2
Summary Statistics for Ideology and Fiscal Adjustments
Our indicator
Ideology LH
Ideology CM
Ideology W
Left
Center
Right
Frequency of Adjustment
12.2
10.8
8.9
Change in Deficit
-2.37
-2.36
-2.34
Change in Spending
-0.74
-1.03
-1.28
Change in Revenues
1.46
1.46
1.09
Frequency of Adjustment
6.6
14.6
12.6
Change in Deficit
-2.18
-2.36
-2.48
Change in Spending
-0.63
-0.70
-1.58
Change in Revenues
1.45
1.59
0.95
Frequency of Adjustment
7.9
14.2
14.8
Change in Deficit
-2.34
-2.14
-2.59
Change in Spending
-0.84
-0.69
-1.67
Change in Revenues
1.37
1.50
0.97
Frequency of Adjustment
8.7
13.9
13.3
Change in Deficit
-2.27
-2.53
-2.34
Change in Spending
-0.78
-1.46
-1.35
Change in Revenues
1.38
1.20
1.09
Note: For each of the four ideology indicators, we present the values for the Frequency of a Fiscal Adjustment and the
changes in Deficit, Spending Revenues for Left, Right and Center cabinets. The variables Change in Deficit, Change in
Spending and Change in Revenues correspond to the yearly changes in the share of the primary public deficit, primary
spending and current revenues as shares of GDP, as defined in Appendix I. Ideology, Ideology LH, Ideology CM and
Ideology W are alternative indices of a cabinet´s ideological orientation.
23
Table 3
Likelihood of Success and Changes in the Public Spending
Probit Estimates – Sample of Adjustments
Dependent Variable: Likelihood of Success
(1)
Total
Spending
(2)
(3)
(4)
(5)
Change In Spending
Public
Public
Public
Consumption Consumption Consumption
Wages
Non-Wages
(6)
(6)
(7)
Transfers
Social
Security
Payments
Subsidies
Public
Investment
Ideology x
Change in Spending
-7.92**
(-2.10)
-23.70**
(-2.00)
-24.35*
(-1.73)
-36.23*
(-1.84)
-0.01
(0.00)
-5.08
(-0.63)
-9.75
(-0.82)
5.75
(0.35)
Ideology
-0.15**
(-2.30)
-0.16**
(-2.11)
-0.12**
(-2.08)
-0.10*
(-1.66)
-0.05
(-1.04)
-0.06
(-1.18)
-0.08*
(-1.69)
-0.05
(-0.81)
Change in Spending
9.74
(0.92)
17.76
(0.56)
17.11
(0.43)
38.26
(0.77)
-23.91
(-0.92)
-15.14
(-0.60)
-14.96
(-0.32)
-89.70*
(-1.72)
Lag Debt
0.30
(0.84)
0.17
(0.55)
0.30
(1.07)
0.15
(0.46)
0.33
(0.96)
0.31
(0.87)
0.28
(0.79)
0.47
(1.40)
Lag Change Debt
1.37**
(2.13)
1.70**
(2.76)
1.22*
(1.76)
1.94**
(2.58)
0.93
(1.34)
1.02
(1.40)
1.28*
(1.77)
1.52**
(2.17)
Majority
-0.40**
(-2.21)
-0.39**
(-2.03)
-0.36*
(-1.93)
-0.40**
(-2.32)
-0.25
(-1.59)
-0.31*
(-1.88)
-0.45**
(-2.57)
-0.44**
(-2.72)
Coalition
-0.35**
(-2.46)
-0.32**
(-2.26)
-0.32**
(-2.30)
-0.35**
(-2.34)
-0.31**
(-2.30)
-0.34**
(-2.45)
-0.36**
(-2.44)
-0.37**
(-2.57)
growth GDP
-1.82
(-0.61)
-2.80
(-0.98)
-3.93
(-1.25)
-2.34
(-0.87)
-4.62
(-1.46)
-4.28
(-1.48)
-3.42
(-1.19)
-0.84
(-0.30)
growth UNR
-0.01**
(-2.37)
-0.01**
(-2.55)
-0.01**
(-2.58)
-0.01**
(-2.69)
-0.01*
(-1.78)
-0.01*
(-1.54)
-0.01**
(-2.41)
-0.02**
(-2.85)
Nr Obs
Log L
Pseudo R2
Observed P
Predicted P
76
-32.62
0.37
0.43
0.33
76
-28.33
0.46
0.43
0.30
76
-30.01
0.42
0.43
0.32
76
-32.84
0.37
0.43
0.36
76
-35.05
0.33
0.43
0.38
76
-34.79
0.33
0.43
0.38
76
-34.09
0.34
0.43
0.38
76
-32.83
0.37
0.43
0.35
Note: ** Significant at the 5 % level and * significant at the 10% level. The independent variable Ideology*Fiscal Change is the product of the
ideological dummy and the yearly change in the variable at the top of each column. For each independent variable we report (dF/dx), i.e., the marginal
change in the probability of success for the average values of the independent variables. In parentheses we report the t-statistic based on robust,
heteroskedastic-consistent standard errors. See Appendix I for a complete description of the variables.
24
Table 4
Likelihood of Success and Changes in Public Revenue
Probit Estimates – Sample of Adjustments
Dependent Variable: Likelihood of Success
(1)
(2)
(3)
(6)
(7)
Total Revenues
Direct Taxes
Social Security
Taxes
Indirect Taxes
Other Taxes
Direct Taxes
Business
Direct Taxes
Households
Ideology x
Change in Revenue
-4.09
(-0.96)
-6.70
(-1.31)
-0.31
(-0.08)
8.45
(0.90)
-34.17**
(-2.69)
-50.99**
(-3.11)
-4.09
(-0.96)
Ideology
-0.03
(-0.56)
-0.05
(-1.14)
-0.06
(-1.20)
-0.09
(-1.35)
-0.02
(-0.47)
-0.05
(-1.19)
-0.03
(-0.56)
Change in Revenue
6.90
(0.50)
-1.80
(-0.11)
12.66
(0.88)
-29.91
(-1.26)
79.31**
(2.39)
132.18**
(2.79)
6.90
(0.50)
Lag Debt
0.37
(1.09)
0.27
(0.85)
0.36
(1.07)
0.32
(0.97)
0.39
(1.24)
0.73**
(1.97)
0.37
(1.09)
Lag Change Debt
1.20*
(1.64)
1.41**
(1.93)
1.37**
(1.89)
1.26*
(1.78)
1.50**
(2.25)
1.58**
(2.17)
1.20**
(1.64)
Majority
-0.38**
(-2.36)
-0.37**
(-2.23)
-0.31**
(-1.92)
-0.35**
(-2.27)
-0.47**
(-2.63)
-0.54**
(-3.32)
-0.38**
(-2.36)
Coalition
-0.33**
(-2.34)
-0.32**
(-2.26)
-0.37**
(-2.65)
-0.33**
(-2.30)
-0.36**
(-2.58)
-0.26*
(-1.86)
-0.33**
(-2.34)
growth GDP
-2.13
(-0.78)
-3.22
(-1.16)
-2.11
(-0.79)
-3.27
(-1.03)
0.14
(0.05)
-0.12
(-0.04)
-2.13
(-0.78)
growth UNR
-0.01**
(-2.38)
-0.01**
(-2.27)
-0.01**
(-2.38)
-0.01**
(-2.50)
-0.01**
(-2.03)
-0.01**
(-2.00)
-0.01**
(-2.38)
Nr Obs
Log L
Pseudo R2
Observed P
Predicted P
76
-33.72
0.35
0.43
0.36
76
-35.44
0.32
0.43
0.37
76
-33.63
0.35
0.43
0.36
76
-35.56
0.32
0.43
0.38
76
-35.31
0.32
0.43
0.38
76
-30.70
0.41
0.43
0.35
72
-25.59
0.48
0.43
0.28
Component
(4)
(5)
Change in Revenue
Note: ** Significant at the 5 % level and * significant at the 10% level. The independent variable Ideo*Fiscal Change is the product of the ideological
dummy and the yearly change in the variable at the top of each column. For each independent variable we report (dF/dx), i.e., the marginal change in
the probability of success for the average values of the independent variables. In parentheses we report the t-statistic based on robust, heteroskedasticconsistent standard errors. See Appendix I for a complete description of the variables.
25
Table 5
Ideology, Change in Spending and Change in Revenues
Probit Estimates – Sample of Adjustments
Dependent Variable: Likelihood of Success
Ideology Indicator
Ideology Ideology Ideology Ideology Ideology Ideology Ideology Ideology
LH
W
CM
LH
W
CM
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Change in Spending
Change in Revenues
Ideology x Change in
Fiscal Variable
-7.92**
(-2.10)
Ideology
-0.15**
(-2.30)
-0.47*
(-1.48)
Change in Fiscal Variable
9.74
(0.92)
Lag Debt
-27.59** -50.40**
(-2.03)
(-3.40)
-6.99
(-0.50)
-4.09**
(-0.96)
-17.12*
(-1.63)
-0.64*
(-1.58)
-16.79*
(-1.73)
-0.69**
(-4.02)
0.05
(0.19)
-0.03
(-0.56)
0.12
(0.76)
0.00
(1.51)
0.24
(1.60)
49.96*
(1.56)
186.82**
(3.26)
-13.03
(-0.29)
6.90
(0.50)
29.32
(1.05)
2.48*
(1.58)
31.63
(1.02)
0.30
(0.84)
0.75*
(1.72)
2.38**
(3.36)
0.69
(1.19)
0.37
(1.09)
0.79*
(1.82)
0.03*
(1.67)
0.85*
(1.67)
Lag Change Debt
1.37**
(2.13)
3.21**
(3.17)
6.20**
(4.07)
3.36**
(2.91)
1.20*
(1.64)
3.01**
(3.25)
0.06*
(1.64)
3.13**
(3.04)
Majority
-0.40**
(-2.21)
-0.35*
(-1.82)
-0.27*
(-1.72)
-0.25
(-1.23)
-0.38**
(-2.36)
-0.38*
(-1.81)
-0.01
(-1.37)
-0.42**
(-1.95)
Coalition
-0.35**
(-2.46)
-0.64**
(-2.24)
-0.99**
(-4.63)
-0.58**
(-2.74)
-0.33**
(-2.34)
-0.57**
(-2.22)
-1.00*
(-1.69)
-0.58**
(-2.44)
growth GDP
-1.82
(-0.61)
1.22
(0.30)
-0.21
(-0.04)
-6.75
(-0.69)
-2.13
(-0.78)
0.65
(0.16)
-0.02
(-0.54)
-1.44
(-0.22)
growth UNR
-0.01**
(-2.37)
-0.03**
(-2.56)
-0.08**
(-3.45)
-0.03**
(-2.11)
-0.01**
(-2.38)
-0.03**
(-2.51)
0.001*
(-1.68)
-0.03**
(-1.95)
Nr Obs
Log L
Pseudo R2
Observed P
Predicted P
76
-32.62
0.37
0.43
0.33
60
-18.05
0.56
0.47
0.36
50
-9.78
0.72
0.46
0.14
55
-14.09
0.63
0.47
0.35
76
-33.72
0.35
0.43
0.36
60
-19.17
0.54
0.47
0.37
50
-7.88
0.77
0.46
0.00
55
-15.59
0.59
0.47
0.32
Note: ** Significant at the 5 % level and * significant at the 10% level. The independent variable Ideo*Fiscal Change is the product of the ideological dummy and
the yearly change in the variable at the top of each column. For each independent variable we report (dF/dx), i.e., the marginal change in the probability of success
for the average values of the independent variables. In parentheses we report the t-statistic based on robust, heteroskedastic-consistent standard errors. See
Appendix I for a complete description of the variables.
26
Table 6
Effect of Changes in Spending and Public Consumption on GDP Growth and Components
Ordinary Least Squares – Sample of Adjustments
Dependent Variable: Growth in GDP, Private Investment and Components, Private Consumption
Dependent Variable
GDP
Private Investment
Business Investment Housing Investment
t to t+1
t to t+2
t to t+1
t to t+2
t to t+1
t to t+2
t to t+1
t to t+2
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
6.98
15.18
5.03
2.93
45.45
59.95
-87.79
-163.48*
Ideology x Change in Public Spending
(0.97)
(1.17)
(0.19)
(0.09)
(1.58)
(1.36)
(-1.41)
(-1.67)
-0.13
-0.18
-0.89
-0.44
-0.72
-0.38
-1.41*
-1.71
Ideology
(-0.95)
(0.26)
(-1.58)
(-0.51)
(-1.07)
(-0.37)
(-1.69)
(-1.23)
-73.06** -154.27** -171.13
-368.11* -358.86** -781.68**
-49.18
164.11
Change in Public Spending
(-2.15)
(-2.79)
(-1.32)
(-1.89)
(-3.35)
(-4.14)
(-0.18)
(0.40)
28.07
54.95
54.27
86.86
11.34
192.37
-31.50
-95.79
Ideology x Change in Public Consumption
(1.39)
(1.37)
(0.77)
(0.76)
(1.41)
(1.37)
(-0.26)
(-0.54)
-0.08
-0.12
-0.73
-0.12
-0.79
-0.24
-0.47
-0.07
Ideology
(-0.56)
(-0.44)
(-1.28)
(-0.13)
(-1.17)
(-0.22)
(-0.58)
(-0.06)
-230.75** -426.23** -704.43** -1445.55** -669.42** -1749.63** -771.44*
-926.64
Change in Public Consumption
(-3.39)
(-2.87)
(-3.09)
(-3.78)
(-2.71)
(-3.73)
(-1.92)
(-1.61)
Private Consumption
t to t+1
t to t+2
(9)
(10)
20.22**
-8.27**
(3.05)
(3.69)
-0.03
-0.12
(-0.22)
(-0.46)
-75.70**
-70.11
(-2.64)
(-1.24)
60.91**
70.26*
(3.36)
(1.89)
-0.01
0.09
(-0.05)
(0.34)
-252.64** -433.68**
(-4.07)
(-2.82)
Note: ** Significant at the 5 % level and * significant at the 10% level. In parentheses we report the t-statistic based on robust, heteroskedastic-consistent standard errors. The dependent
variables are noted in the header row and include the growth rate of GDP and the change in each of its components, namely Private Investment, Business Investment, Housing Investment and Private
Consumption. For each dependent variable we use the accumulated rates of change, one and two years after the fiscal adjustment. Values are in percent of GDP for components of GDP and in growth
rates for GDP itself. Here we report only the coefficient of the ideology-fiscal policy interaction, the ideology and the change in fiscal variable, but all specifications include the control variables in
Tables 3 and 4, (not reported here for reasons of parsimony).